International Journal of Mineral Processmg, 38 ( 1993 ) 157-175
157
Elsevaer Science Publishers B V , A m s t e r d a m
Influence of grinding method on complex sulphide ore flotation: a pilot plant study K.S. Eric Forssberg a, T.V. S u b r a h m a n y a m b* a n d L e i f K. N i l s s o n a ~Dzvtston of Mmeral Ppoce~smg, Lulea Untverstt)' of Technology, S-951 87 Lulea, Sweden bDepartamento de Geologta, CCE/UFRN, Campu~ Umversltarw, Catxa Postal 1576, 59072-970 Natal-RN, Brazil ( Received 11 February 1992, accepted after revaslon 16 November 1992 )
ABSTRACT The present work deals with pilot plant investigations on the influence of conventional- and autogenous grinding methods m the flotation of a complex sulphide ore_ The study Includes the investigation of the particle size and shape analysLs and the pulp chemical measurements - - E,, pH, O2, concentration of cations and anions, at several stages of the grinding and flotation orcmts The rougher flotation concentrates of the Cu-Pb and Zn circuits were analysed to evaluate the flotation behavlour of chalcopynte, galena and sphalerlte The results indicate better hberatlon of minerals m coarse size fractions for the pulp ground by autogenous mill while conventional grinding shows higher degree of hberat~on m the fine size range Metallurgical analyses of the final flotation concentrates shows higher recoveries ofCu, Pb and Zn for the pulp ground by autogenous mdhng. The grades of Cu and Pb were better with conventional grinding but the recoveries were less than those obtained by aulogenous grinding Several interesting observations concerning pulp chemlstr~ (e_g the concentration of Fe ions is lower m pulps ground b~, conventional method than by autogenous means ~) are d~scussed
INTRODUCTION
Ore milling ahead of flotation is an important step to liberate the valuable mineral particles from the gangue. In sulphide mineral fotation the method of grinding adopted influences the pulp chemistry and consequently the flotation behaviour of the mineral particles. Depending on the milling environment, i.e. reducing or non-reducing, several reactions take place. Redox reactions, dissolution o f minerals, speciation, mineral-reagent interactions, precipitation, adsorption of the reaction products on mineral surfaces, etc. are a result of the interactions between the minerals, the reagents and the type of grinding environment. The effects of these interactions are well understood w~th single mineral systems on a laboratory level. But studies on pilot plantor industrial scale with complex ores are less common. A multitude of varl*Corresponding author_
0301-7516/93/$06 00 © 1993 Elsevier Science Pubhshers B V All rights reserved
158
k S E FORSSBERG ET AL
ables, some controllable, some not, govern the process and any step to optimxze the conditions to improve the plant performance warrants an overall understanding of those interacting variables. The present work reports the results of pdot plant investigations on the influence of conventional and autogenous grinding methods in the flotation of a complex sulphide ore. The study includes the particle size and shape analyses and the pulp chemical measurements, pH, Eh, O2, concentration of cations and anions, at several stages of the grinding and flotation circuits. EXPERIMENTAL
Test material The complex sulphide ore was from Renstrom, Boliden (Sweden), containlng chalcopyrite, galena, sphalerite and pyrite together with gold and silver minerals. Due to a long storage time the ore was found to have suffered some oxidation. The feed (0-20 m m ) was prepared by crushing and care was taken to obtain a homogenous material for both the circmts 1.e. conventional and autogenous grinding, but some slight differences were noticed. Table 1 shows the average grades of feed to the flotation circuit.
Grmdmg ctrcutts Figures 1 and 2 show the flow charts of the grinding and flotation circuits. The flotation scheme adopted was the same for both grinding methods The difference between the flow charts is m the grinding circuits.
Conventional grinding The dimensions of the rod and ball mills used for primary and secondary grinding were the same: diameter 1.32 m and length 2.5 m. A cyclone (inner diameter. 140 m m ) was in circuit and the cyclone overflow was the feed to the flotation circmt with a feed rate of 2.0_+ 0.1 t/hr. The size of the feed to the primary mill was <25 mm. The details of sample collection points are shown in Fig. 1. TABLE 1 Grades of feed to flotation circuit (_ lrcmt
Cu
Pb
Zn
S
Au
~g
Convenllonal 3~ulogenous
0 62 0 58
1 47 1 76
6 47 7 5l
12 00 13 50
1 40 I 40
134 154
INFLUENCE OF GRINDING METHOD ON COMPLEX SULPHIDE ORE FLOTATION
A~
faC~
Sample collecuon Capacity measurement Steveanalysts Pulpchemlstry Fracuonanalysts ~ Z ~
C S P F
ORE
099m3/h
C PF ~t '
V ~ J CYCLONE IV ] Na 2C0~280 g./t[
P,S t 194't/h 67 wt %
159
) ~====[j 19;2 [P~2 ROD MILL 55 wt % [ =
155mJ/h[
~tP
/ i,............ J ~] 6 8 w t % Ig_~_~fi'i'~J [ • BALL MILL I
3 4 CuPb FLOTATION
1 94 t/h 4 2 w t %
-
2 85 m
3/h
CONDITIONER 048t/h 397 w t %
105mJ/h
t
0 22 t/h 17 6 wt %
~263 t/h )5 492m3/h
1 02 rn J/h NaHSOj 145 g/t]
rough conc 1 0 10 t/h
- []'6
~
0Cu-Pb 11006335m3/h t/hc°nc wt %
2 3~t/h
3 80 m 3/h
GRADES OF rough cons 1 rough cone 2
1 88 t/h 406wt%
75m ~/h
Zn FLOTATION 297 t/h 640 w t %
4 42 m J/h
Cu Pb Zn
Ca(OH) 2 2000 g/t
49 217 178
46 164 187
. ¢ ~ j ICuSO, 18o g/'l
~
3 65 t/h 60 0 wt c~
8 51 t/h
2 44 rn J/h
9 61 m 3/h rough cone 1 0 63 t/h
Zn cone 0 lgt/h 335wI~
0 36 m 3/h 1 70 t/h ~ A 4 2 6 9 wt %
62 m J/h TAILINGS
GRADES OF rough cons 1 rough conc 2 (%) (%) Zn 41 1 37 9 Cu 04 05 Pb 18 17
Fig 1 C o n v e n t i o n a l g r i n d i n g / f l o t a t i o n c~rcult
Autogenous grinding Primary grinding was carried out in a Hardinge mill, the dimensions of which were: diameter 3 0 m and length 0.915 m. The mill discharge was directly conveyed to a 3 m m sieve, the undersize of which was fed to the cyclone (tuner diameter: 140 m m ) . The overflow of the cyclone was the feed to the flotation circuit while the underflow was circulated to the secondary pebble mill (diameter 1.32 m and length 2.5 m ) . Both the mills were lined with rubber and it must be pointed out that the pebble mill endplates were made of
160
K S E FORSSBERG ET AL
A
~
x C S P
Capacny measurement Sieveanalysts
Samplecollection
F
Fr~;Uon analysJs [ ' - 1
Pulpchemlstry ~
ORE
,
1 0 6 t/h
l 29 rn 3/h
M
[-~1' CYCLONE
MILL ~1: I
41-47 wt %
C, P, F
I
I
I
• ,,, ~BLE~ILI~ I 3
4
25t/h 44wt%
-
318m3/h
CONDITIONER ~ . . . n 4 14 t/h
t
67 wt % lf~-~'~--~d
~p
0 59 t/h 30-32 wt % 1 3 8 m 3/h
, 5 5 84 m ~/h
NaH~
rough conc 1 0 15 t/h
125 g/l]
~
Cu-Pb conc 0 12 t/h 334wt%
'6
0 24 m 3/h
3 44 t/h 5 91 m 3/h
GRADES OF roughconc 2 (%) 18 28 8 250
rough conc 1
38 t/h 30-38 wt %
q•2
62 m 3/h
Zn FLOTATION 223 t/h 537 wt% 1 92 rn J/h
Cu Pb Zn
C~(OH) 21700 g/II
(%) 21,1 32 1 21 1
2 18 t/h 43-53 wt %
6 79 t]h
~r7
8 90 m 3/h roughconc 1 0 93 t/h
L 236mJ/h
Zn cone 0 35 t/h 24-25 wt % (18 6 wt%)
'8 4 26 t/h
1 04 m J/h
7 70 m 3/h
2 03 t/h ~ , A 5 2 5 - 2 7 wt % 78rn3/h
TAILINGS
GRADES OF roughconc 1 roughconc 2 (%) (%) Zn 419 386 Cu 04 04 Pb I9 19
F~g 2 A u t o g e n o u s g r i n d i n g / f l o t a t i o n c l r c w t
steel. The pebble mill was charged with pebble media of 35-60 mm. Feed rate measured was 2.4 _+0.1 t/hr. The sample collection points are shown in Fig. 2. Flotation clrcull
Figures 1 and 2 show the flotation schemes together with sample collection points in bold numbers ( 1 to 8 ). The results of the pulp chemical analysis are shown in Figs. 10-14, wherein steps 1-8 correspond to the following sample collection points: 1. primary mill outflow; 2 cyclone underflow; 3. secondary
INFLUENCEOF GRINDING METHOD ON COMPLEXSULPHIDEORE FLOTATION
| 6|
mill outflow; 4. cyclone overflow; 5. feed to Cu-Pb flotation; 6. rougher tails to Cu-Pb scavenger flotation; 7. feed to zinc flotation and 8. zinc rougher tails to scavenger flotation. Table 2 gives the reagent schedules. The analysis of feed water used in both grinding and flotation circuits xs shown in Table 3 TABLE2 Reagents added to Cu-Pb and Zn flotation c,rcults Objective
Reagents
Dosages m g/t conventional*
pH reg in Cu-Pb clrcmt (pH 8 5) Cu-Pb flotation
For Zn depression in C u - P b flotn pH reg m Zn circuit (pH 12 0) For zinc flotation Zinc actwahon Frother (Zn circuit)
sodmm carbonate Aeroflot 404 (mercaptobenzoth~ozole) Aeroflot 31 (dlth~ophosphate) Ke 6266 (dnsopropyld~th~ophosphate ) sodium blsulph~te zinc sulphate Yarmor-F calcium hydroxide ~sobutyl xanthate copper sulphate Yarmor-F
autogenous
280
240
43
35
4
3
43
41
145 315 2 2000
125 270 2 1700
90 180 3
80 170
Yarmor-F= Pine Oil *Reagent dosages are higher m comparison to autogenous grinding circmt due to a h~gher proportion of fines generated in the conventional grinding cxrcmt
TABLE 3 Analysis of water pH Eh CO2 02 C1Ca 2+
SO~$20~Fe~o,al Pb 2+ Cu 2+ Zn 2+
7 50 + 457 mV 2 0),( 10 -4 M 16 0 mg/l 1.14 mg/1 7 00 mg/1 7 00 mg/1 0 O0 mg/l 0 25 mg/I 0 00 mg/l 0_03 rag/1 0.19 mg/I
162
K S E FORSSBERG ET 4.L
METHODS
Degree ofhberatlon and particle shape determinations The liberation size of particles was determined by a petrographic microscope. The test samples were s~eved into different size fractions and polished sections were prepared. The number of gangue, mineral and associated particles was determined in each size fraction. For particle shape analysis a Jeol 733 Super Electron Scanning Microscope with back-scattered electron xmage analyser was used. The analysis was carried out w~th a LINK AN 10000 Mxcro Analyser System and the products analysed were: the flotation feed and the concentrates of C u - P b and Zn. In every sample 30-40 picture fields were analysed at 150 to 500 times magnification. The number of particles measured was 1500-2000 in each sample. For each particle 20 diameters were measured. Particle shape was determined by dividing the smallest Feret diameter by the largest Feret diameter. L~kew~se, 10 classes of part|cle shape were separated, 1.e. the smallest being 0-0.1, corresponding to needle shape, and the largest 0.9-1.0, being Isometric particles.
Pulp chemlstry studtes The parameters mesured to determine the chemical conditions in the pulp were: redox potentml, pH, oxygen concentration, concentrations of Ca, Fe, Cu, Pb, Zn, sulphur anions and total sulphur (Stot). Pulp samples for the determinatxons of metal ions and S~ot were filtered through a 0.22 ~m membrane. Sulphur anions - - thiosulphate, sulphate and sulphate, were analysed by ion chromatography. The S,o, of the pulp was determined by the combustion method. For the analysis of thiosulphate, sulphite and sulphate, the samples were collected directly from the flotation cell with a special sampling syringe consisting of standard millipore equipment (Forssberg and Halhn, 1989). Sufficient care was taken to avoid oxidation of the ions and hence the time taken did not exceed 5 min, from the sample collection to the injection of the sample into the analyser. A Dionex 2010i ion chromatograph was used for the analysis of sulphur anions. The equ|pment consists of a column for anion analysis and two detectors---one electrochemical and another, conductometric. The two detectors differ in their sensitivity to different ions. Thiosulphate was determined with the electrochemical detector and sulphite and sulphate, with the conductivity detector. The redox potentials and pH were measured directly in the pulp. pH was determined by a Radiometer P H M 83T, with a GK 2410 C combination electrode. The redox potential was measured with a platinum electrode P 101 to-
163
INFLUENCEOF GRINDING METHODON COMPLEXSULPHIDEORE FLOTATION
gether with a K 401 calomel electrode, both connected to a Radiometer PHM 52b pH meter. Before each measurement the pH meter contacts were cleaned by careful polishing with a moistened paper cloth strewn with fine grained A1203 powder. The oxygen concentration of the samples was measured by a GIMAT OXI 74 FM oxygen meter with a OXI 722 Z1 electrode equipped with a builtin automatic temperature compensator. The metal 1on concentrations were determined by Atomic Absorption Spectroscopy. RESULTS
Degree of liberation and shape factor The volume percentage of free particles in the flotation feed is shown in Table 4. In - 5 3 + 38 and > 53/~m size fractions higher degrees of liberation are observed with autogenous grinding for chalcoyrite, galena and pyrite, while conventional grinding seems to be better in fine size range ( - 3 8 + 20/.tm). In general it may be pointed out that autogenous grinding gives better liberation. In fine size range conventional milling shows higher degrees of liberation since the proportion of fines produced is higher. The results of the analysis of shape factor are shown in Figs. 3 and 4 for the flotation feeds of conventional and autogenous grinding circuits. The difference m particle shapes of the products produced by the two methods of grinding was not significant excepting that the particles were more isometric m the pulp produced by autogenous grinding. A large proportion of particles have shape factors in the range 0.4-0.7, while only a very small fraction corresponds to shape factors 0-0. l and 0.9-1.0.
Flotation The overflow fraction of the cyclone was fed to the flotation circmt and Fig. 5 shows the particle size distribution in the flotation feed. The flow chart TABLE 4 Volume percentage of free particles in flotation feed Mineral
Chalcopynte Galena Sphalente Pynle
Conventional grinding
Autogenous grinding
20-38pm
38-53#m
>53pm
20-38/zm
38-53/xm
> 53gm
99 100 85 99
69 78 69 95
63 20 61 57
93 93 79 99
82 80 68 98
78 67 60 76
164
K S E FORSSBERG
ET AL
• 20 - 38 pm t ~ 38 - 53 pm
25
20
[] +53pm
~ - n ~ L
o 10
Z 01
02
03
04
05 06 07 Shape factor
08
09
10
Fig. 3 Shape factor of particles m the flotatton feed (conventional grinding) 25
I I 20 - 38 pare
~0
[ ] 38- 53 pm [[] + 5 3 0 m
'8 10
0
Ol
02
03
04
06 ' 07 ' 08 05 Shape factor
09
I0
F~g 4. Shape factor of parlicles m the flotation feed (autogenous grinding)
Flokltton feed
10~3 9C
Conventional ql Autogcnous
0
80
~
5C
20
ol
I 10
I 20
I 30
I 40
I I I I I I 50 60 80 100
Particle size (pro)
Ftg_ 5. Particle size distribution in flotation feed.
consists of two flotation circuits----one for C u - P b and the other, for Zn flotation. Each flotation circuit consists of two rougher cells (the concentrates of which will be referred as rougher flotation concentrate 1 and 2 ), one scavenger which receives the tailings from the rougher cells 1 and 2; the rougher flotation concentrate 1 was conveyed directly to a final re-cleaner cell while rougher flotation concentrate 2 was circulated to a cleaning step. The cleaner
INFLUENCE
OF GRINDING
METHOD
ON COMPLEX
SULPHIDE
ORE FLOTATION
165
concentrate was then subjected to a final re-cleaning. The float products of the re-cleaner cells were the final concentrates in the C u - P b circuit and the Zn flotation circuits, respectively. The tailings of the re-cleaner cell were recycled to the cleaner cell. The tailings from the scavenger cell of the C u - P b circuit, rich in Zn content, were conveyed to the Zn flotation circuit. Other details such as the circulating loads, pulp densities, etc., are shown in Figs. 1 and 2 and the reagent dosages are given in Table 2. The flotation behav~our of chalcopynte, galena and sphalerite was evaluated in the rougher concentrates of C u - P b and Zn circuits. The weight percentage of each size fraction ranging from < 10 to 98/lm, was determined in the flotation feed and in the rougher flotation concentrates, together w~th the percentages of chalcopyrite, galena and sphalerite in the respective s~ze fractions. The mineral recoveries were calculated from the respective masses present in the feed and m the rougher flotation concentrates. For the calculation of mineral recoveries in rougher concentrate 2, the masses of minerals floated m rougher concentrate 1 were subtracted from the masses of minerals present in the flotation feed and the rest was considered to be the feed for rougher flotation 2. The recoveries of minerals as a function of average particle size are shown in Figs. 6-9. In general galena shows a high floatabllity both in conventional and autogenously ground pulps. The recoveries of chalcopyrite and sphalerite, however, vary depending on the particle size (Fig. 6). The difference in the floatabilities of galena for the ore ground by conventional and autogenous circuits becomes clear upon examining F~g. 7. In that, the recoveries of both galena and sphalerite (autogenous) increase with increasing particle size up to ~ 30/~m, and then decrease. The recovery of chalcopyrite (autogenous) is m i n i m u m where galena and sphalerlte (autogenous) show m a x i m u m floata-
Cu-Pb ClrLult rougher flotation conc 1 10(3 90 80
CuFeS 2 PbS ZnS
Conventional
Aulogenous
o x A
• + •
70 ~
60
~ so ~
40 30 20
10
20
30
40
50 60
go 100
Average particle size/prn)
Fig_ 6. M i n e r a l recoveries tn r o u g h e r flotation c o n c e n t r a t e 1 ( C u - P b o r c u l t )
166
K S E FORSSBERG
ET AL
Cu-Pb clrcmt rougher flotauon conc 2 CuFeS 2 PbS ZnS
ConventtonaJ O
Autogenous •
I00
90 8(1 7O 60 50 "
40 30 2(I 10 [ 2(I
0
I M)
I 40
I I I I I I 50 6(1 80 l(13
A ~erage par lit le *,lie I ~lll
Fig 7 Mineral recoveries in rougher flotation concentrate 2 ( C u - P b circuit)
Z n circuit rougher flotation cone 1 I(KI 9O /40
g
CuFeS 2 PbS ZnS
Con~enuona]
Autogenous
0 x ~
• + •
70
60 f, 5O s~ 40 3O 2(I ID I(I
20
30
4.verage p,lrlltle
stte
40
50 60
I I I g0 10¢)
(pro)
Fig. 8 Mineral recoveries in rougher flotatmn concentrate 1 (Zn clrcmt)
bihtles and the trend is the same in both the rougher flotation concentrates of Cu-Pb circuit (Figs. 6 and 7). Galena and chalcopyrite (conventional) exhibit almost a similar flotation behavlour (Figs. 6 and 7). In the cleaning and re-cleaning steps of Cu-Pb circuit, ZnSO4 and NaHSO4, were added, respectively, to depress sphalerite and hence the reason for low recoveries of sphalerite in rougher flotation concentrates of Cu-Pb circuit (Figs. 6 and 7). Figures 8 and 9 show the recoveries of minerals in rougher concentrate 1 and 2, respectively, of the Zn circuit. Sphalerite was activated by the addition of CuSO4 and the pH was regulated to 12 with Ca(OH)2 (see Figs. 1 and 2). Sphalerite (autogenous) shows higher flotation recoveries in rougher concentrate 1 (Fig. 8) and this trend is reversed in the rougher concentrate 2 (Fig. 9), where maximum
INFLUENCE OF GRINDING
METHOD
167
ON COMPLEX SULPHIDE ORE FLOTATION
Zn ca-cult rougher flotation cone 2 Con',entlonal CuFeS 2 PbS ZnS
I(X)
Autogenous
0
•
x A
+ •
A
90
80 70
e~
~
6O
4U
30 2O 10 0
I 10
I 20
I 30
I 40
I I 50 60
I I I I ~;0 100
A~ erage partitle MTe (pm)
Fig. 9 Mineral recoveries m rougher flotatxon concentrate 2 (Zn circuit)
TABLE 5 Metallurgical data of concentrates Circuit/concentrate
~t%
grade % Cu
Conventlonal/Cu-Pb conc_, Autogenous/Cu-Pb Cone, Conventlonal/Zn conc, ~utogenous/Zn conc,
3 4 9 13
09 80 38 92
16 10 0 0
70 80 20 30
recover) %
Pb
Zn
Cu
Pb
Zn
30 70 27 30 1 50 I 80
4 60 7 90 55_40 49 40
82 00 86 00 4 00 7 00
69 00 76 00 l 1 00 14 00
2 5 85 92
00 00 00 00
floatability is exhibited by sphalerite (conventional). The flotation recoveries of sphalente, galena and chalcopyrite ground by conventional and autogenous methods show quite opposite trends (Fig. 9). Table 5 gives the grades and recoveries of final concentrates of Cu-Pb and Zn and Table 6 shows the volume percentage of free particles in those concentrates. In the autogenous Cu-Pb concentrate the volume percentage of free galena floated was poor in coarse size fractions ( - 53 + 38 and > 53 # m ) even though the percentage liberated was higher in the feed (Table 6). Particle shape analysis of the flotation concentrates revealed that the floatability of particles having shape factors of 0.4-0.8 were higher, the highest being 0.50.6. In other words, particles of intermediate shapes i.e. neither needle nor spherical, but with ratios of the shortest and the longest Feret diameters of 2 : 3 or l : 2, floated much better.
168
K S E F O R S S B E R G E T AL
TABLE 6
Volume percentage of free parhcles m flotation concentrates Product
orcmt
Cu-Pb con Zn con
galena
chalcopynte
(a)
(b)
(c)
cony autog con~
91 91 100
autog
100
86 85 100 100
84 80 88 67
(a)
sphalente (b)
(c)
83 76 93
73 36 -
17 24 -
100
-
100
pyrite
(a)
(b)
(c)
(a)
(b)
(c)
88 94 93 92
75 97 86 86
56 80 84 85
97 100 100 100
100 100 100 100
100 lO0 100 100
(a) 20-38 pm, ( b ) 38-53 lzm, (c) > 5 3 p m
Pulp chemistry In pulp chemical studies the average value of three readings measured for each variable was considered and the results are shown in Figs. l 0-14. Figure Convenuon~ O pH A Eh Q 02
Eh 0 2 (mV) (mg/I) 3~-
Autogenous • pH • Eh • 02
pH
-13
330-
-12
3(~-
"ll
270
~0
240
15
"9
210
lO
.8
I0
180 150
7 1
2
3
4
5 6 Steps
7
6
8
Fig 10. Chemical condlUons at different stages of grinding/flotation orcult
4(]0-
Conventional O S~ A Rest S
AuLogcnous • S total • Rest S
350 300-
~
250-
150-
50 2 3 4
5 6 7 Steps
6
Fxg 11. Concentratxon of sulphur
INFLUENCE OF GRINDING
Conventlonnl
so~
0 A V
METHOD ON COMPLEX SULPHIDE ORE FLOTATION
| 69
Autogenous
SO4 S 203 SO 3
• • •
SO4 S 203 SO3
$ 2 0 3z (m/g) 110
(rag/I) 9OO
9(1
800-
80
700 60050O
-
4(10
50 4(I
~O0 . 2O0 lo0
20 -
d.
1
2
3
4 5 ~leps
6
7
8
Fig 12 Concentration of sulphur anions
Convenuonal
400-
Autogcnous
o Ca
•
Ca
350 300-
% 250 E 2OO150 10050-
2
3
4 5 Sleps
6
7
8
Fig 13 Concentration of calcium ions
10 shows the Eh, 02, and pH at different stages. In Eh measurements a calomel electrode was used as a reference and therefore, the actual value was added to + 252.8 mV. Eh and pH variations are a result of either reagent additions or other reactions occurring at different stages of the circuit. The E h variations in conventional and autogenous circuits, as observed from Fig. 10, are close to each other. When the pulp is transferred to the flotation cell the E h rises due to stirring and dilution of the pulp and hence the observed trend (steps 5 to 6; Fig. 10). A subsequent fall in the Eh value (step 6 to 7) was due to the addition o f h m e (pH 12) to the zinc flotation circuit (see Figs. 1 and 2 ). The oxygen levels do not show any significant difference betwen the autogenous and conventional circuits, excepting that the 02 concentration is slightly higher in the autogenous mill (step l; Fig. 10). Figure I 1 shows the total sulphur and rest sulphur concentrations, where
170
K S E FORSSBERG
03
0 A V Q
Fe
Cu Pb Zn Fe
• • • •
Zn
04
Autogenous
Conventional
Cu Pb Zn Fe
ET AL
03
o2 0!
Ol
0
I
[
I
[
[
I
I
I
I
I
I
I
I
I
I
[
1
2
3
4 5 ~leps
6
7
8
1
2
3
4 5 ~teps
6
7
8
A 2
A 3
A A 4 5 Sltp~
6
7
8
Pb
02
20
001
~ 10
t
0 t
I 2
I 3
I I 4 5 Step~
I 6
I 7
I 6
T
0 /
Fig 14 Concentrationsof metal ions
the latter was calculated by deducting the sulphur originating from the anions - - $ 2 0 ~ - , SO3- and SO4- from the total sulphur. A higher total sulphur content in the primary mill outflow (step l ; Fig. l 1 ) may be due to a higher proportion of sulphide fines produced by conventional grinding in comparison to autogenous milling. A subsequent decrease in the total sulphur concentration (steps 1 to 3) may be due to the removal of fines by the cyclone and/ or the effect of dilution (Na2CO3 for pH regulation). In further steps ( 3 - 6 ) the total sulphur concentration continuously increases alike both In conventional and autogenous circuits due to secondary milling. The autogenous circuit shows a slightly higher concentration of sulphur (steps 4-6; Fig. 1 1 ) since the circulating load or charge was more than the conventional circuit (see Figs. I and 2). The concentration of sulphur a n l o n s - - S 2 0 3 - , SO~- and SO4- are shown in Fig. 12. A high SO24- concentration in the pulp ground by conventional method (step 1 ) does not indicate oxidation during grinding but on the other hand, as already pointed out, due to long storing time the ore must have suffered oxidation. Wearing of oxidized surfaces i.e. intense cleaning action when the ore was subjected to conventional grinding, and consequent release of SO4-, might result in a higher concentration. Autogenous grinding affects the grain boundaries deeper but does not clean the oxidized surfaces as in conventional grinding and hence the reason for a lower SO~- concentration in the pulp ground by autogenous milling (step 1 ; Fig. 12). In further steps there is no significant change in the concentration of SO42-. The thiosulphate content in the pulps ground by conventional and autogenous methods shows fluc-
I N F L U E N C E OF G R I N D I N G M E T H O D ON COMPLEX S U L P H I D E ORE FLOT~.TION
17 1
tuatlons, the latter being more significant than the former. The S 2- may oxi&ze according to, 2S2- + 3/20~_--,$2042- + 2 e -
(1)
and in an autogenous mill the environment can be expected to be more oxi&zing (or non-reducing) than in a conventional mill and therefore, a higher $203- concentration. It may be argued that long storing time must have caused the formation ofthiosulphate, as is the case with SO4 . In such a case a higher $203- content should also be observed in the pulp ground by conventional m e t h o d (due to the grinding mechanism described earher) which is not the situation, as observed in Fig. 12. The quantity of the ore treated in the autogenous circuit was higher in comparison to that of the conventional circuit and such a difference may also contribute significantly to a higher $203- concentration in the autogenous circuit. About 1-5% of dolomite (corresponding to 20-100 k g / h r ) was associated with the ore and was the source for Ca 2+, Mg 2+ , and C O 3 - . Ca 2+ ions also originate from the addition of CaC12 to the ore to avoid freezing (Figs. 1 and 2). Since dolomite is a soft mineral, a higher proportion of fines will be produced in conventional grinding circuit than in autogenous grinding. Therefore these factors must have resulted in high Ca :+ concentrations in the primary grinding mill of the conventional circuit (step 1; Fig. 13); whereas a lower Ca 2+ concentration in the autogenous mill may be due to the lower a m o u n t of fines. Due to the addition of Na2CO3 to the secondary mill for pH regulation (see Figs. 1 and 2) excess Ca 2+ may react and precipitate in the form of CaCO3, which in turn may result in a lower Ca 2+ concentration, as observed in Fig. 13 (step 3 ). The secondary mill discharge product analysed 5-10 mg Ca2+/l. The solubility of CaCO3 is 14 m g / l which corresponds to a Ca 2÷ concentration of 6 mg/1 In further steps there is a uniform increase in Ca 2+ followed by a sudden rise (step 6 to 7) which was due to the addition of h m e to the zinc circuit. From Figs. 1 and 2 it can be observed that the a m o u n t of C a ( O H ) 2 added to the conventional circuit was higher (2000 g/ t) than that was added to the autogenous circuit (1700 g / t ) , and hence a higher Ca :+ concentration in the former (step 7; Fig. 13 ). However, step 7 to 8 (conventional) shows a decrease in the Ca 2÷ concentration while that in the autogenous circuit registers a slight increase, and finally both attain almost the same level of concentration (step 8 ). A decrease in the Ca 2÷ concentration m the conventional circuit (step 7 to 8) may be due to the effect of dilution caused by the differences in the circulating loads of the pulp in the conventional and autogenous circuits (note the difference in the quantities of rougher concentrates of Zn circuits; Figs. 1 and 2). Figure 14 shows the metal ion concentrations at different stages of the grinding and flotation circuits. The iron concentration in the pulp ground by conventional m e t h o d was expected to give a higher value in comparison to
172
K S E FORSSBERG ET AL
the pulp ground by autogenous mill. The situation in reality was found to be contrary i.e. the concentration of iron measured in the pulps ground by conventional and autogenous methods were 0.04 and 0.09 mg/1, respectively (step 1; Fig. 14). When these values are compared with the iron present in the mill water (Table 3; Fe~toL)= 0.25 mg/1 ), it becomes clear that the iron concentration decreases upon grinding. The reason for such a behaviour could be that the iron In the solution may react and form oxy-hydroxide precipitates of iron and coat the mineral surfaces. Whereas such reactions may be less prominent m an autogenous mill and hence a higher iron concentration in comparison to the pulp ground in the conventional mill. The CuSO4 added to the Zn flotation circuit for the activation of Zn surfaces follows the chemical exchange reaction, Cu2+ +ZnS--*CuS +Zn2+
(2)
and a strong affinity of Cu 2÷ for sphalerlte results in the release of Zn ions thus leading to higher Zn concentrations (step 6 to 7; Fig. 14). The Zn ion concentration (conventional) is higher while lower Zn content in the autogenous circuit can be expected to be due to the formation of Zn oxide and its precipitation. Copper ion concentration does not show any rise as the Zn ion since Cu -~+ is consumed according to the reaction shown in Eq. (2) Pb ion concentration also shows a rise which can be explained by, ZnS + Pb 2+ o ( Z n ) - P b S + Zn 2+
(3)
and ( Zn )-PbS + Cu: + o ( Z n ) - C u S + Pb 2+
(4 )
~.e. Pb ions react with the zinc mineral surface and a subsequent activation by Cu 2÷ ions may lead to the release of Pb ions (step 6-7; Fig. 14). However, the uptake of lead ions by sphalerite is low in comparison to copper ions. DISCUSSION
Particle size reduction occurs due to the breaking forces which are applied on particles in different directions. Depending upon the grinding method, abrasive wear, impact wear, erosion and corrosion, occur during grinding. The conventional and autogenous methods differ in their mechanisms of exercising forces on the particles during comminution. The balls and rods used as the grinding media in conventional grinding exercise a point and line load, respectively, on the particles. In autogenous grinding the grain boundaries are affected and the proportion o f fines produced, in general, is lower in comparison to conventional grinding. The influence of grinding on the floatabihty of some sulphide minerals was studied in terms o f galvanic Interactions (Rey and Formanek, 1960; A d a m et
INFLUENCE OF GRINDING METHOD ON COMPLEX SULPHIDE ORE FLOTATION
173
al., 1984; Rao and Natarajan, 1990). The type of grinding influences the surface chemistry of the sulphide mineral and consequently its flotation behavlour (Rao et al., 1976; Berglund and Forssberg, 1987; Forssberg et al., 1988 ). These aspects are dealt with in greater detail in recent publications (Rao and Natarajan, 1991; Subrahmanyam and Forssberg, 1992 ). In the wet grinding of sulphides in a conventional mill the mineral-grinding media interactions play a dominant role since the particles are always in contact with the grinding media (iron) which has a lower rest potential than the sulphide minerals. The sulphide fines produced, due to a large cathodic surface area, accelerate the anodlc oxidation of the media. The dissolved iron ions produced as a result of the anodic oxidation of the grinding media react with the hydroxyl ions generated by the cathodic reduction of oxygen, thus forming hydroxy complexes of iron, which are hydrophdic, and may coat the mineral surfaces. Such coatings may impair the floatability of the sulphide mineral by interfering with xanthate adsorption as observed by Rao and Natarajan (1990) with chalcopynte. A reducing environment in conventional grinding may be due to any of the following reactions (Guy and Trahar, 1984), Fe 2+ + 2 e - = F e
(Eo = - 0.440V)
Fe(OH)2 + 2 H + + 2 e - = F e + 2 H , O Fe304 -1-8H + -.I-8e- = 3Fe + 4H20
(5) (Eo = - 0.047v) (E,, = - 0 . 0 8 4 V
)
(6) (7)
The presence of iron essentially governs the type of environment m a grinding mill. Several factors like the pulp density, sulphide mineral content and their reactiwty, aeration effects in the equipment used, are responsible for oxygen levels. A recent investigation (Houot and Duhamet, 1990) emphasizes the importance of oxygenation of pulps in the flotation of sulphide ores. Oxygen is quickly consumed in situations where new sulphide surfaces are continuously formed and also due to the abraded iron during grinding. Iron may be present in nearly all flotation feeds, or even otherwise, may originate from the mill components like for example, in the present investigation, from the autogenous mill side walls and this is suspected to be one of the reasons for almost similar redox potentials in both the circuits (Fig. 10). In an autogenous or a prcelain mill, the type of e n w r o n m e n t is non-reducing and generally a higher potential can be expected. In oxidizing environment sulphide minerals react, M e S = M e 2+ +S 2-
(8)
2S 2- + 3H20 = $203- + 6H + + 8e-
(9)
$20~- + 5 H 2 0 = 2 S O 2 + 1 0 H + + 8 e
(10)
and both thlosulphate and sulphate ions are formed depending on the time
174
K S E FORSSBERG ET AL
and the type of environment. However, higher thiosulphate concentrations in the pulp ground by autogenous mill in comparison to conventional (Fig. 12 ) indicate that the environment was non-reducing. High thiosulphate concentration was found to form coatings of insoluble salts on mineral surfaces (Jonsson, 1983). Such coatings were considered to be responsible for the depression of pyrite ground in a porcelain mill (Berglund and Forssberg, 1987). Although no data is available on the flotation behaviour of pyrite in the present study, some factors which may be responsible for pyrite depression may be mentioned: the collector, dlthiophosphate, used in the C u - P b circuit does not form dithiophosphatogen (the species responsible for pyrite flotation) at pH > 4. Further, at high pH vlues (as in the Zn circuit) the pyrite suface may behave like that of a ferric hydroxide. In the C u - P b flotation circuit, sphalerlte was depressed by the addition of ZnSO4 and NaHSO3 to the cleaner and re-cleaner cells, respectively. In the Zn flotation circuit, the Ca 2+ ions originating from the addition of Ca(OH)~ favour pyrite depression. Similar to the mechanism of pyrite depression at high pH's, the hydroxyl ions depress the flotation of chalcopyrite in the Zn circuit The depression of galena in the Zn circuit occurs, however, due to the formation of plumbite Pb (OH) 3-. The flotation of sphalerite by longer chain xanthate, like lsobutyl, is well known and further the Zn surfaces were activated by the addition of CuSO4. From the foregoing, it becomes clear that grinding of sulphide minerals in a ferruginous environment (conventional) may lead to the formation of hydroxy complexes which in turn coat the sulphide mineral particles and ~mpair the floatabllity. In an autogenous mill, due to the non-reducing environment, the thiosulphate salts formed may coat the surfaces of pyrite and depress its floatabllity. Pyrite, almost invariably, is associated with many sulphides. A better selectivity in flotation for the ore ground by autogenous mill probably is one of the reasons why it is preferred in practice. CONCLUSIONS
The results of the investigations show better liberation of particles in coarse size fractions for autogenous grinding. In fine size ranges, the degree of liberanon was better with conventional grinding. The particle shape analysis does not show any significant difference between the two methods of grinding. The reactions occurring during grinding depend upon the milling environment i.e. reducing or non-reducing. The presence or absence of iron in the milling environment has an important role in the nature of reactions occurring in a mill In a reducing environment the hydroxy complexes formed may coat the sulphide mineral surfaces and impair the floatability while in an oxidizing environment, the insoluble thiosulphate salts may coat the minerals, for example pyrite, and may improve the selectivity towards the valuable sulphide
INFLUENCEOF GRINDING METHODON COMPLEXSULPHIDEORE FLOTATION
17 5
minerals. Metallurgical data of the final concentrates shows higher recoveries of Cu, Pb and Zn for the pulp treated by autogenous grinding and was at the cost of their grades. The grades of Cu and Pb were better with conventional grinding but the recoveries were less than those obtained by autogenous grinding. An important aspect that remains to be addressed is that the differences observed between the two mehtods of grinding on a pilot scale are only marginal (according to the metallurgical data!) and are not as marked as the results that are generally obtained under controlled conditions in the laboratory and several hypothesis may be used to explain such variations. ACKNOWLEDGEMENTS
Financial support from the Swedish National Board for Technical Development and Boliden Mineral AB is gratefully acknowledged. The contributions of Messrs. Thierry Dumas and Tommy Nordqvist are acknowledged and finally, a special thanks goes to Dr. Nlls Johan Bolin for reviewing the manuscript.
REFERENCES Adam, K , Natarajan, K A. and lwasakl, I , 1984_ Grinding media wear and its effect on the f l o t a u o n o f s u l f i d e m l n e r a l s Int J Miner Process, 12 39-54_ Berglund, G_ and Forssberg, K S E , 1987 Pulp chemlstr~ measurements in sulphide mineral flotauon Erzmetall, 40(4) 189-195 Forssberg, K S E and Halhn, M I , 1989. Process v~ater reclrculatlon tn a lead-zinc plant and other sulphide flotation plants In K V_S Sastry and M C Fuerstenau (Edflors), Challenges m Mineral Processing, SME AIME, Ch. 27, pp 452- 466 Forssberg, K S_E, Sundberg, S and Hongxm, Z , 1988 Influence of different grinding methods on floatablhty. Int J Miner Process, 22 183-192 Guy, P J and Trahar, W.J., 1984 The Influence of grinding and flotauon environments on the laboratory batchflotation of galena Int J Miner Process, 12 15-38_ Houot, R and Duhamet, D., 1990 Importance of oxygenation of pulps In the flotat|on of sulfide ores Int J Miner Process, 29 77-87 Jonsson, H , 1983_ Progr_ Rep. No 2, Mlnfo Rep No 1501 (Swedish Mineral Processing Research Foundation, Stockholm) (as referred by Berglund and Forssberg, 1987) Rao, M_K Y and Natarajan, K A_, 1990_ Studies on chalcopynte ore grinding with respect to ball wear and effect in flotation Miner Metall Process, 35-37 Rao, M K Y and Natarajan, K_A, 1991 Factors Influencing ball wear and flotation with respect to ore grinding Miner Process Extract Metal[ R e v , 7 137-173 Rao, S R , Moon, K S and Leja, J_, 1976 Effect of grinding media on the surface reactions and flotation of heavy metal sulphldes. In M_C. Fuerstenau (Editor), Flotation A_M Gaudln Memorial Volume SME AIME, Vol 1, Ch 17, pp 509-527 Rey, M and Formanek, V., 1960 Some factors affecting the selectivity m the differential flotanon of lead-zinc ores in the presence of oxidized lead minerals In Proc V Int Miner Process Congr_ lnst_ Mln Metall, London, pp 343-353 Subrahmanyam, T V. and Forssberg, K S E , 1992 Mineral/solution interface chemistry in minerals engineering In. Minerals Engineering, 92, Vancouver In press